On Wed, May 27, 2020 at 6:43 PM Mark Adams <mfad...@lbl.gov> wrote: > Nvidias's NSight with 2D Q3 and bs=10. (attached). > > I am using LU in serial. > > I copied MatSetValues_SeqAIJ into a .cu file and made some minor > adjustments. > > I am getting the indices from Matt's DMPlexMatGetClosureIndices and the > matrix is from DMPlex. I also use GPU direct solvers sometimes and will > want that in the future. All and all I figure AIJ is a safer bet, but maybe > BAIJ is an option. > > Matt: Could I use BAIJ with Plex? >
Plex does this automatically if you have blocks. Matt > Thanks, > Mark > > > On Wed, May 27, 2020 at 6:14 PM Jed Brown <j...@jedbrown.org> wrote: > >> What did you profile to determine that expanding indices is significant? >> >> matsetvaluesblocked4_ was made specially for PETSc-FUN3D with BAIJ >> matrices. >> >> I take it you can't use BAIJ because you use GAMG? >> >> Mark Adams <mfad...@lbl.gov> writes: >> >> > Is there a Mat AIJSeq method For MatSetValuesBlocked, >> > like matsetvaluesblocked4_ that is not hardwired for bs=4? >> > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
